Talin 1 and Paxillin Facilitate Distinct Steps in Rapid VLA-4-mediated Adhesion Strengthening to Vascular Cell Adhesion Molecule 1*

VLA-4 (α4β1) is a key integrin in lymphocytes, interacting with endothelial vascular cell adhesion molecule 1 (VCAM-1) on blood vessels and stroma. To dissect the contribution of the two cytoskeletal VLA-4 adaptor partners paxillin and talin to VLA-4 adhesiveness, we transiently knocked them down in Jurkat T cells and primary resting human T cells by small interfering RNA silencing. Paxillin was required for VLA-4 adhesiveness to low density VCAM-1 under shear stress conditions and was found to control mechanical stability of bonds mediated by the α4 subunit but did not affect the integrin affinity or avidity to VCAM-1 in shear-free conditions. Talin 1 maintained VLA-4 in a high affinity conformation, thereby promoting rapid VLA-4 adhesion strengthening to VCAM-1 under both shear stress and shear-free conditions. Talin 1, but not paxillin, was required for VLA-4 to undergo optimal stimulation by the prototypic chemokine, CXCL12, under shear stress conditions. Interestingly, talin 1 and paxillin played the same distinct roles in VLA-4 adhesions of primary T lymphocytes, although VLA-4 affinity to VCAM-1 was at least 200-fold lower in these cells than in Jurkat cells. Collectively, our results suggest that whereas paxillin is a mechanical regulator of VLA-4 bonds generated in the absence of chemokine signals and low VCAM-1 occupancy, talin 1 is a versatile VLA-4 affinity regulator implicated in both spontaneous and chemokine-triggered rapid adhesions to VCAM-1.

VLA-4 (␣ 4 ␤ 1 ) is a key integrin in lymphocytes, interacting with endothelial vascular cell adhesion molecule 1 (VCAM-1) on blood vessels and stroma. To dissect the contribution of the two cytoskeletal VLA-4 adaptor partners paxillin and talin to VLA-4 adhesiveness, we transiently knocked them down in Jurkat T cells and primary resting human T cells by small interfering RNA silencing. Paxillin was required for VLA-4 adhesiveness to low density VCAM-1 under shear stress conditions and was found to control mechanical stability of bonds mediated by the ␣ 4 subunit but did not affect the integrin affinity or avidity to VCAM-1 in shear-free conditions. Talin 1 maintained VLA-4 in a high affinity conformation, thereby promoting rapid VLA-4 adhesion strengthening to VCAM-1 under both shear stress and shear-free conditions. Talin 1, but not paxillin, was required for VLA-4 to undergo optimal stimulation by the prototypic chemokine, CXCL12, under shear stress conditions. Interestingly, talin 1 and paxillin played the same distinct roles in VLA-4 adhesions of primary T lymphocytes, although VLA-4 affinity to VCAM-1 was at least 200-fold lower in these cells than in Jurkat cells. Collectively, our results suggest that whereas paxillin is a mechanical regulator of VLA-4 bonds generated in the absence of chemokine signals and low VCAM-1 occupancy, talin 1 is a versatile VLA-4 affinity regulator implicated in both spontaneous and chemokine-triggered rapid adhesions to VCAM-1.
A remarkable feature of leukocyte integrins is their ability to modulate their adhesive activity within fractions of seconds through cytoplasmic associations with cytoskeletal partners (1,2). These associations may regulate both integrin conformation (affinity) and clustering prior to or immediately following ligand binding and thereby increase integrin avidity to surfacebound ligands. Leukocyte integrins like VLA-4 and LFA-1 must recognize their endothelial ligands and stabilize adhesion under constant shear forces. Recent evidence suggests that when the integrin headpiece is occupied by ligand, it can be instantaneously stabilized by application of external force (3), so shear forces exerted on leukocytes at target sites on blood vessels may play both positive and negative roles in the adhesive outcome of integrin-ligand interactions. To load forces, integrins and ligands may need to be properly anchored to a rigid surface, such as the cortical cytoskeleton (4). Specific cytoskeletal partners of integrins may therefore regulate not only the affinity and clustering states of their integrin targets but also the efficiency by which an integrin occupied with ligand can strengthen its bond under the influence of external forces. In support of this possibility, we recently showed that the adhesive activity under shear of the key immune cell integrin, VLA-4, is modulated by an ␣ 4 cytoplasmic association with paxillin, without any noticeable changes in its affinity, clustering, or binding avidity to VCAM-1 2 measured in the absence of shear (5). In that study, an ␣ 4 tail mutant, Y991A, with disrupted binding to paxillin and reduced recruitment of talin 1 to the ␤ 1 subunit of the integrin heterodimer, exhibited reduced VLA-4 adhesiveness to VCAM-1 measured under shear stress conditions (5). Since the adhesive defect of the integrin mutant could result from impairments in both paxillin and talin associations with the VLA-4 heterodimer, the specific contribution of each of these two key cytoskeletal integrin linkers to stabilization of VLA-4-VCAM-1 interactions under shear flow remained obscure.
Talin 1, the key talin member in immune cells, is an antiparallel actin-binding homodimer that serves as a major link between integrin and the actin cytoskeleton and is involved in the early steps of focal adhesion assembly (6,7). The binding of the talin head region to the ␤ subunits of multiple ␤ 2 and ␤ 3 integrins breaks a suppressory clasp, enabling the conformational change of the integrin ectodomains from bent (inactive) states into extended states, with partially or fully active headpieces (8,9). However, conformational modulation of ␤ 1 integrins by intact talin has not been demonstrated to date, although different inside-out signals can recruit talin to integrin sites via the Ras-related integrin-activating GTPase, Rap-1 (10).
Paxillin is a 68-kDa multidomain adaptor protein that integrates signals from integrins and growth factor receptors to actin cytoskeletal remodeling effectors that coordinate cell motility and turnover of focal adhesions (11). In lymphoid cells, although integrin-mediated focal adhesions are not generated, paxillin plays a key role in ␣ 4 -mediated motility via high affinity * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  binding to the ␣ 4 integrin tail (12,13). The ␣ 4 integrin-paxillin complex regulates the activation of FAK and Pyk2, deactivates Rac in the leading edge of migrating cells (14), and can also trans-activate the LFA-1 integrin in T cells (15). These and other functions are tightly regulated by multiple tyrosine and serine/threonine phosphorylation sites on paxillin (16,17). So far, a role for paxillin in rapid integrin-mediated T cell adhesions has not been explored. In T cells, VLA-4 interactions with VCAM-1 can generate stable shear-resistant adhesive contacts within subseconds even without a requirement for prior rolling contacts (18,19). In an earlier study, we observed that suppression of paxillin and talin each reduced the adhesiveness of VLA-4 in cells settled for 1 min on VCAM-1 before being subjected to shear stress (5). Physiological interactions between these counterreceptors at the vasculature takes place under continuous shear flow and must generate adhesion strengthening within seconds. Therefore, and in light of the key involvement of talin and paxillin in the much slower organization of integrin assemblies within static contacts (20), the contribution of these adaptors to the earliest VLA-4 interactions with VCAM-1 under continuous shear could be masked by these later events.
In this study, we identified distinct roles for paxillin and talin 1 in the control of VLA-4 adhesiveness to VCAM-1 developed at adhesive contacts over a time frame of seconds and subjected to continuous shear stress. Using selective siRNA-targeted suppression of talin 1 and paxillin in a Jurkat T cell model, we found that paxillin, although not regulating VLA-4 affinity or avidity to VCAM-1 in shear-free systems, was essential for mechanical stabilization of individual VLA-4/VCAM-1 bonds at subsecond contacts subjected to shear flow. Talin 1, on the other hand, was implicated in direct modulation of VLA-4 affinity to VCAM-1 and consequently in VLA-4/VCAM-1 adhesions as well as in LFA-1 adhesion. Talin 1, but not paxillin, also contributed to in situ chemokine induction of VLA-4 adhesiveness to VCAM-1. This study identifies talin 1 and paxillin as specialized modulators of distinct adhesive properties of VLA-4 essential for rapid VLA-4 adhesiveness developed under shear forces.
Immunofluorescence Flow Cytometry-Cells were washed once with cation-free H/H medium (HBSS containing 2 mg/ml bovine serum albumin and 10 mM HEPES, pH 7.4), resuspended in binding medium (H/H medium supplemented with 1 mM CaCl 2 and 1 mM MgCl 2 ), and incubated with primary antibodies (10 g/ml) for 30 min at 4°C. The samples were then washed and incubated with secondary antibodies for 30 min at 4°C. Cells were washed and analyzed immediately on a FACScan flow cytometer (Becton Dickinson, Erembodegem, Belgium). To probe the basal and agonist-stimulated expression of the activation reporter epitopes 15/7 and HUTS21, cells were incubated with the mAbs (10 g/ml) in binding medium for 5 min at 37°C in the presence or absence of agonist (CXCL12 (30 nM) or phorbol 12-myristate 13-acetate (100 nM)), followed by a 25-min incubation on ice to minimize integrin internalization. Cells were then washed and incubated with secondary antibodies for an additional 30 min at 4°C.
siRNA-mediated Silencing of Paxillin and Talin 1-Silencing of talin expression in Jurkat cells was achieved by a single talin 1-specific 21-nucleotide siRNA (Dharmacon, Lafayette, CO) corresponding to positions 6043-6063 (5Ј-AAUCGUGA-GGGUACUGAAACU-3Ј) relative to the talin 1 mRNA start codon (26) screened from several talin-specific oligonucleotides. Silencing of paxillin was performed using a single siRNA duplex sequence (5Ј-CCCUGACGAAAGAGAAGC-CUAUU-3Ј and 5Ј-UAGGCUUCUCUUUCGUCAGGGUU-3Ј), as described (14). Control transfections were performed with a fluorescein-labeled 21-nucleotide duplex directed to luciferase GL2. Transfection of both Jurkat T cells and primary PBLs was carried out by electroporation using the Nucleofection system (Amaxa). Transfected cells were maintained in culture medium. Talin and paxillin expression, monitored by immunoblotting, was maximally suppressed at 72 and 96 h post-transfection in Jurkat cells or human PBLs, respectively, the time points chosen for subsequent functional assays.
Western Blot Analyses-For Western blot studies, 1 ϫ 10 7 cells (treated or untreated) were solubilized in 100 l of lysis buffer (5) and 20 l of lysates were separated by SDS-PAGE in reducing buffer. Blots were developed using enhanced chemiluminescence (Sigma).

VCAM-1 Microbead Binding
Assay-Protein A-coated magnetic Dynabeads (M-280; Dynal Biotech Inc., Lake Success, NY) were coated at room temperature with various concentrations of VCAM-Fc in H/H binding medium, washed according to the manufacturer's instructions, and stored on ice. Cells and VCAM-1-coated beads were mixed at room temperature for 1 min in binding medium at a concentration of 1 ϫ 10 7 cells/ml at a ratio of 1 cell to 8 beads, followed by a 3-fold dilution in binding medium. The cellular side scatter, distinguishing between bead-bound and bead-free cells, was analyzed immediately in a FACScan flow cytometer (Becton Dickinson). Background binding determined with Protein A-coated beads was Ͻ5% of the maximal binding observed at VCAM-1 saturation and was subtracted from the total binding results.
Soluble VCAM-1-Fc Binding Assay-Cells (0.5 ϫ 10 6 ) were incubated with (VCAM-1) 2 -Ig (VCAM-1-Fc) for 30 min at room temperature in H/H binding medium. Cells were then washed twice by suspension in H/H binding medium and incubated with phycoerythrindonkey anti-human IgG (Jackson ImmunoResearch Laboratories) for 20 min at room temperature (in binding medium) before washing with PBS and FACS analysis. For each cell population, the relative cell binding at a given concentration of soluble VCAM-1 was deduced by the fluorescence level of cell-bound VCAM-1 relative to the maximal cell-bound VCAM-1. For Scatchard analysis of the binding data, absolute values of maximal VCAM-1 binding (B max ) to various siRNAtreated cells were derived by analysis of the maximal cell-bound VCAM-1-Fc content, present at saturation, which was removed by EDTA and quantified by enzymelinked immunosorbent assay. To assess the effect of transient chemokine stimulation on VLA-4 affinity to VCAM-1, cells were incubated at room temperature with increasing concentrations of VCAM-1-Fc for 25 min, and a fixed concentration of the chemokine of interest was added for an additional 5-min FIGURE 1. Paxillin is crucial for spontaneous VLA-4-mediated tethers to low density VCAM-1. A, Jurkat cells were transiently transfected with paxillin specific or control luciferase siRNA. Total lysates were immunoblotted with paxillin or total ERK-specific mAbs. Densitometric analysis revealed a decrease of 80% in paxillin content. B, FACS staining of ␣ 4 (left) and ␤ 1 (right) of control and paxillin RNA interference-transfected cells. The thin black line depicts isotype-matched control. C, frequency and strength of interactions of Jurkat cells transfected with control (Cont.) or paxillin (Pax.) siRNA with low, medium, or high density sVCAM-1 at a shear stress of 0.75 dyne/cm 2 . The mean ϩ range of two independent experiments is depicted. Results are representative of four independent experiments. D, PBLs were transiently transfected with paxillin-specific or control luciferase siRNA, and total lysates of each group were immunoblotted with mAb to paxillin and total ERK. Densitometric analysis revealed a decrease of 60% in paxillin content. E, tethering, frequency, and strength of binding by PBLs transfected with control or paxillin siRNA to the indicated densities of sVCAM-1 at a shear stress of 0.75 dyne/cm 2 . The mean ϩ range of two independent experiments is depicted. Results are representative of three independent experiments. F, FACS staining of L-selectin on Jurkat-treated cells. The thin black line depicts isotype-matched control. G, frequency and strength of tethers formed by Jurkat cells transfected with control or paxillin siRNA with peripheral node addressin (PNAd) at a shear stress of 1 dyne/cm 2 . The mean ϩ range of two fields in the same experiment is depicted. Results are representative of two independent experiments. period. Cells were immediately fixed with 2% paraformaldehyde for an additional 10 min and washed. The relative VCAM-1 binding was assessed by immunostaining as described above.
Laminar Flow Adhesion Assays-Purified ligands or mAbs were coated alone or with CXCL12 on polystyrene plates as previously described (25). Site densities of coated sVCAM-1 were determined as described (25,27). The polystyrene plates were each assembled on the lower wall of the flow chamber (260-m gap) (28). Cells were washed with cation-free H/H medium, resuspended in binding medium (H/H medium supplemented with 1 mM CaCl 2 and 1 mM MgCl 2 ), and perfused through the flow chamber at the desired shear stress. All flow experiments were conducted at 37°C. Tethers were defined as "transient" if cells attached briefly (Ͻ2 s) to the substrate and as "arrests" if they immediately arrested and remained stationary during at least 3 s of continuous flow. Frequencies of adhesive categories within differently pretreated cells or rates of cell accumulation on adhesive substrates were determined as a percentage of cells flowing immediately over the substrates, as previously described (25). Over 95% of cellular tethers to VCAM-1 were blocked by pretreating cells with the specific VLA-4 blocker, BIO-1211 (21) (1 g/ml), and all tethers to ICAM-1 were blocked in the presence of EDTA.

Paxillin Is Required for VLA-4 Tether Formation under Shear Stress but Not for Integrin Avidity to VCAM-1 under Shear-free
Conditions-To examine the role of paxillin in T cell VLA-4 adhesiveness to VCAM-1 in T cells in adhesive contacts under shear stress, we knocked down paxillin expression in Jurkat cells by small interfering RNA (siRNA)-mediated silencing. Suppressing paxillin expression by up to 80% using transient siRNA silencing (Fig. 1A) did not affect integrin ␣ 4 or ␤ 1 expression ( Fig. 1B) but resulted in dramatically reduced VLA-4mediated attachment of the T cells to both low and medium VCAM-1 densities under continuous shear flow (Fig. 1C). This paxillin-mediated suppression of VLA-4 tethers was also evident at higher shear stresses (data not shown). Identical paxillin suppression did not affect, however, the frequency or strength of VLA-4-mediated Jurkat attachments to high density VCAM-1 (Fig. 1C). These results were confirmed with multiple Jurkat clones expressing different levels of VLA-4. Consistent with the Jurkat data, paxillin suppression in PBLs (Fig. 1D) retained VLA-4 expression (not shown), partially inhibited VLA-4-mediated attachment to medium density VCAM-1, and had no effect on initial attachments or on their strength of interaction to high density VCAM-1 (Fig. 1E). Since PBLs express 3-5-fold lower levels of VLA-4 than Jurkat T cells, their attachments to low density VCAM-1 were undetectable even without paxillin suppression (not shown).
The inhibitory effects of paxillin suppression on VLA-4mediated attachments, adhesive events generated within subsecond contacts with the VCAM-1-coated surface, could result from a general modulation of the number and function of lymphocyte microvilli. We therefore next tested whether L-selectin adhesiveness, which strongly depends on microvilli and cytoskeletal integrity (29 -31), was also inhibited by paxillin suppression in Jurkat cells. Since paxillinsuppressed cells retained L-selectin adhesiveness under various shear stress conditions (Fig. 1, F and G), we conclude that paxillin suppression in Jurkat T cells did not globally alter their shape, cytoskeletal integrity, or microvillar structure. Rather, paxillin affected VLA-4 adhesiveness by interfering with the ability of the integrin to form adhesive tethers at subsecond contacts, especially to low density VCAM-1.
A failure of an integrin to form tethers could result from a reduced affinity to ligand, impaired clustering, and/or an inability of the tether to resist disruptive forces (5). We therefore next considered that paxillin suppression may directly reduce VLA-4 avidity to surface-bound VCAM-1 under shear free conditions. However, VLA-4-mediated Jurkat cell binding to VCAM-1-coated beads was not affected by paxillin suppression at any density of VCAM-1 tested (Fig. 2). Paxillin suppression also did not affect the binding of soluble VCAM-1 to VLA-4 on Jurkat cells (data not shown), further suggesting that neither affinity nor avidity of VLA-4 to surface-bound VCAM-1 are regulated by paxillin in Jurkat T cells. Thus, the adhesive defect of paxillin-suppressed cells reflected a mechanical rather than a direct binding defect of the VLA-4-VCAM-1 bond.

Lck-dependent Tyrosine Phosphorylation of Paxillin Is Not Required for Its Contribution to VLA-4 Adhesiveness under
Shear-Paxillin function is modulated by multiple tyrosine phosphorylation sites (11). In order to determine whether paxillin requires tyrosine phosphorylation to mechanically stabilize VLA-4-VCAM-1 bonds, we used the Lck-deficient cell line, JCAM1.6, which expresses normal levels of VLA-4 but is defective in constitutive Lck-mediated tyrosine phosphorylation of numerous target proteins, including paxillin (32). Paxillin tyrosine phosphorylation was confirmed to be completely abrogated in the Lck-deficient cells (Fig. 3A), supporting an exclusive role for the Lck kinase in this phosphorylation. The adhesiveness of VLA-4 in these Lck-deficient cells, although lower than in WT cells (28), was still highly sensitive to knockdown of paxillin (Fig. 3C), although the degree of paxillin silencing in these cells was lower than in normal Jurkat cells (Fig. 3B versus Fig. 1A). Thus, even in the absence of Lck-dependent tyrosine phosphorylation of paxillin, paxillin is still essential for optimal VLA-4 adhesiveness to VCAM-1 under shear stress.
Talin 1 Contributes to VLA-4 Adhesiveness under Shear Stress Conditions by Stabilizing VLA-4 at a High Affinity State-To address how talin 1 contributes to VLA-4 adhesiveness to VCAM-1 under shear stress conditions, we next analyzed the effect of siRNA-mediated silencing of talin 1, the exclusive talin member on T cells, on VLA-4-mediated cell attachments to VCAM-1 under shear flow. Talin 1 suppression (80%) (Fig. 4A) did not alter ␣ 4 and ␤ 1 expression (Fig. 4A) but reduced VLA-4 tether formation to both low and high VCAM-1 densities, with only minimal effect on tether formation to medium density VCAM-1 (Fig. 4B). Talin suppression also reduced the fraction of tethers that firmly arrested the attached (tethered) cells on both medium and high density VCAM-1 (Fig. 4B). As with paxillin suppression, talin 1 suppression did not affect L-selectin-mediated attachments to the selectin ligand, peripheral node addressin (data not shown), ruling out a global effect of the suppression on lymphocyte shape and microvillar function.
Notably, unlike paxillin suppression, talin 1 suppression dramatically reduced VLA-4 avidity to VCAM-1 coated on beads under shear free conditions (Fig. 5A). Talin 1 suppression also attenuated the direct equilibrium binding of soluble VCAM-1-Ig to Jurkat T cells (Fig. 5B), suggesting that the major effect of talin 1 suppression on VLA-4 avidity to VCAM-1 is a reduction in VLA-4 affinity. Scatchard analysis of the binding data between VCAM-1-Ig and control siRNA-treated Jurkat VLA-4 yielded a high affinity subset that binds VCAM-1-Ig with a K d of 5 nM and comprises one-fifth of the total VCAM-1-Ig binding sites on these cells (Fig. 5B) (data not shown). Talin 1 suppression reduced the affinity of this subset by 7-fold without significantly reducing the overall number of VCAM-1 binding sites (Fig. 5B) (data not shown). Consistent with a key role of talin in VLA-4 activation via the ␤ 1 subunit of the integrin, talin suppression,  but not paxillin suppression, also significantly reduced the fraction of the ␤ 1 integrin subunit stabilized in an activated conformation probed with the 15/7 reporter mAb (Fig. 5C) (data not shown). We thus conclude that talin 1 contributes to the stabilization of activated VLA-4 with high affinity to VCAM-1 necessary for optimal T cell adhesiveness to surface-bound VCAM-1 both under shear stress and shear-free conditions.
To further dissect how talin 1-stabilized high affinity VLA-4 contributes to VLA-4-mediated cell attachment to high density VCAM-1, we selectively blocked either high affinity VLA-4 or the total functional VLA-4, respectively, in the presence of low (1 ng/ml) and high (1 g/ml) concentrations of the small molecular weight VLA-4 ligand, Bio1211 (Fig. 6). Selective blockage of high affinity VLA-4 with low concentration Bio1211 specifically abrogated a subset of tethers that immediately arrested untreated T cells on the VCAM-1 without impairing weak VLA-4 mediated tethers (Fig. 6), lasting less than 1 s (data not shown) and without suppressing rolling interactions. These transient and rolling tethers were completely blocked by a high concentration of Bio1211. These findings suggest that rolling and transient tethers are probably mediated by intermediate and low affinity VLA-4 subsets, respectively. Notably, VLA-4 on talin 1-suppressed Jurkat cells was completely insensitive to low concentrations of Bio1211, consistent with the notion that these cells lost all high affinity VLA-4 binding sites (Fig. 6). Nevertheless, the intermediate and low affinity VLA-4 remaining in talin 1-suppressed Jurkat cells were sufficient to mediate capture with reduced frequency of firm tethers (Fig. 4B). These experiments collectively suggest that talin 1 is required for a subset of high affinity VLA-4 to engage with immobilized VCAM-1 during subsecond contacts. This high affinity subset plays a key role in the attachment of intact Jurkat cells mainly to low density VCAM-1 as well as in the stabilization of firm tethers that immediately arrest cells on medium and high density VCAM-1 (Fig. 4B).

VLA-4 Affinity to VCAM-1 on Primary T Lymphocytes Is Negligible, but Spontaneous VLA-4 Adhesiveness Is Critically
Dependent on Talin 1-VLA-4 on primary T cells is expressed at lower levels than on Jurkat T cells, and therefore higher VCAM-1 densities are required for lymphocytes to interact with the endothelial ligand (28). Nevertheless, talin 1 suppression (60%), which did not alter ␣ 4 or ␤ 1 expression (not shown) resulted in inhibitory effects on VLA-4-mediated tethering and firm adhesions, which were comparable in extent with those observed in Jurkat cells (Fig. 4, B and C). Notably, VLA-4 on primary T cells failed to bind soluble VCAM-1 even at a concentration that saturated Jurkat VLA-4 (Fig. 5B). Conversely, VLA-4 on Jurkat could bind VCAM-1 at concentrations at least 200-fold lower than that which failed to bind T cell VLA-4 ( Fig.  5B) (data not shown). Consistent with this, ␤ 1 integrin conformation on resting T lymphocytes was largely inactive, as probed by the activation reporter mAb, 15/7 (Fig. 5C, right). Thus, although VLA-4 in primary T lymphocytes is stabilized in an inactive state with negligible affinity to VCAM-1 under shearfree conditions, under shear stress conditions, this VLA-4 not only readily interacts with VCAM-1 but is as dependent on talin 1 as Jurkat VLA-4.  Paxillin, but Not Talin 1, Augments Shear Resistance of ␣ 4 Tethers to a Surface-bound ␣ 4 mAb-In addition to generating high affinity VLA-4, talin 1 is a cross-linker of adjacent integrins (33) and may therefore be involved in affinity-independent integrin adhesiveness. We next tested the effect of talin 1 suppression on the intrinsic attachment capacity of the ␣ 4 subunit within VLA-4 to surface immobilized ␣ 4 mAb, HP1/2, under shear stress conditions. This mAb was shown by us to similarly recognize both low and high affinity VLA-4 (28). Suppression of paxillin inhibited VLA-4-mediated capture to the ␣ 4 -specific mAb, particularly at low and medium densities (Fig. 7A) (data not shown). A similar role for paxillin was observed in primary resting T lymphocytes on high density ␣ 4 mAb (Fig. 7C). Thus, paxillin, via its direct association with the ␣ 4 tail, plays a key role in an affinity-independent ability of the integrin to generate shear resistant tethers when occupied with a surface-bound mAb. Furthermore, the role of paxillin is conserved in resting T cells irrespective of the relative affinity of their VLA-4 to VCAM-1. In contrast, and consistent with its specialized role in affinity regulation of VLA-4, talin 1 suppression had no affect on ␣ 4 attachment to the HP1/2 mAb at subsecond contacts in Jurkat or primary T lymphocytes (Fig. 7, B and C) or at longer lived adhesive contacts (data not shown), regardless of the ␣ 4 mAb density tested (Fig. 7B) (data not shown). Thus, paxillin rather than talin 1 regulates mechanical stabilization of ligated ␣ 4 under shear stress conditions. Talin 1 Is Implicated in Rapid Activation of VLA-4 by the Chemokine CXCL12-Chemokines can activate both VLA-4 and LFA-1 on tethered lymphocytes via distinct activation mechanisms (34). Talin 1 suppression in PBLs interferes with chemokine-stimulated inside-out LFA-1 activation and with rapid chemokine triggered LFA-1-mediated arrest on ICAM-1 (25,26). We next considered the possibility that talin 1, paxillin, or both may play a role in rapid chemokine-facilitated VLA-4 activation under shear stress conditions. Interestingly, talin 1 suppression in both Jurkat T cells and primary lymphocytes impaired CXCL12-stimulated VLA-4 adhesiveness to VCAM-1 triggered in situ by the chemokine at subsecond-lived contacts, whereas paxillin did not (Fig. 8, A-C). Furthermore, talin 1 but not paxillin contributed to VLA-4 activation by a second prototypic PBL chemokine, CCL21 (data not shown). Thus, paxillin is not involved in VLA-4 activation by in situ chemokine-triggered signals, in contrast to talin 1. Notably, chemokine signals failed to switch VLA-4 from low to high activation conformation, as detected by the reporter mAbs, 15/7 (25) (Fig.  8D) and HUTS21 (data not shown), suggesting that in the absence of ligand, chemokine-triggered VLA-4 stimulation under shear stress conditions does not involve an inside-out conformational switch of the integrin. Furthermore, despite its robust stimulation of VLA-4 adhesiveness to VCAM-1 under shear stress (Fig. 8B), CXCL12 failed to increase VLA-4 affinity to VCAM-1 measured under shear-free conditions (Fig. 8E). These results therefore implicate talin 1 in rapid VLA-4 activation by CXCL12 and VCAM-1 signals at short lived contacts.
Paxillin Is Not Required for LFA-1 Adhesiveness to ICAM-1-Paxillin associates with the ␣ 4 integrin subunit tail but not with other ␣ integrin subunits (12). In light of studies reporting a cross-talk between VLA-4 and LFA-1 (35), we next wished to verify that the effects of paxillin on VLA-4 adhesiveness are specific and do not result in any indirect effects on LFA-1 adhesiveness to ICAM-1 at subsecond contacts as well as during prolonged (1-min) contacts. As expected from previous results with PBLs (26), talin 1 suppression retained LFA-1 expression (Fig. 9A) but inhibited LFA-1-mediated attachments to ICAM-1 and abrogated the ability of LFA-1 to develop shear FIGURE 7. Paxillin, but not talin 1, augments shear resistance of ␣ 4 integrin tethers to ␣ 4 mAb independent of VLA-4 affinity states. A, reduced tethering and firm adhesion to low (1 g/ml) and medium (5 g/ml) density immobilized anti-␣ 4 mAb (HP1/2) mediated by Jurkat cells transfected with paxillin siRNA (Pax.) compared with cells transfected with control siRNA (Cont.). B, similar tethering and firm adhesion to immobilized anti ␣ 4 mAb (HP1/2) mediated by Jurkat cells transfected with either control or talin 1 (Tal.) siRNA. C, tethering and firm adhesion to immobilized anti ␣ 4 mAb (HP1/2, 10 g/ml) mediated by PBLs transfected with either control or talin 1 or paxillin siRNAs. In A-C, the mean ϩ range of two independent experiments is depicted. Results in A-C are representative of three independent experiments. resistant adhesions in settled cells (Fig. 9B). Paxillin silencing, on the other hand, had no effect on LFA-1-mediated Jurkat attachments to ICAM-1 under shear flow (Fig. 9C), adhesive interactions that develop on a subsecond time scale. In addition, paxillin knockdown did not affect the ability of LFA-1 to generate shear-resistant adhesions in cells settled for 1 min on ICAM-1 before being subjected to shear flow (data not shown).
Thus, paxillin is not involved in LFA-1-mediated adhesiveness, whereas talin 1 is critical for both LFA-1 and VLA-4 adhesiveness developed under diverse dynamic conditions.

DISCUSSION
This study demonstrates that paxillin and talin 1 facilitate distinct aspects of VLA-4-mediated adhesion strengthening on VCAM-1 under shear flow (Table 1). Paxillin is required for VLA-4 adhesiveness to low density VCAM-1 solely under shear flow and in the absence of potent chemokine signals but is dispensable for VLA-4-mediated adhesions under shear-free conditions. Paxillin determines the ability of VLA-4 to promote initial lymphocyte attachments to low density VCAM-1, consistent with its role in conferring VLA-4 bonds increased resistance to immediate rupture by shear stress. Paxillin is not implicated in VLA-4 affinity to ligand nor for activation of VLA-4 by chemokine signals. Thus, at high VCAM-1 densities or when sufficient chemokine-activated VLA-4 molecules are stabilized in situ at the adhesive contact, the rupture force exerted on each VLA-4-VCAM-1 bond is lowered, and the contribution of paxillin to cell adhesion decreases proportionally. Paxillin has recently been shown to interact preferentially with high affinity ␣ 4 integrins (36). Nevertheless, paxillin contributed to VLA-4 adhesiveness also in PBL despite the low affinity of the integrin in these lymphocytes. In sharp contrast, talin 1 is essential for the generation of a VLA-4 subset with high affinity to VCAM-1, and it thereby contributes to VLA-4-mediated adhesions under both shear stress and shear-free conditions. Talin 1 also facilitates VLA-4 activation by chemokine signals. Thus, talin 1 is a versatile and general integrin affinity regulator in T cells, whereas paxillin is a specialized mechanical regulator of VLA-4-VCAM-1 bond stability and resistance to rupture by external shear forces.
How do paxillin-␣ 4 associations confer the resistance to strain of VLA-4-VCAM-1 bonds? Anchorage provides both selectins and integrins with an ability to interact with their ligands under external strain and to generate shear-resistant adhesiveness (5,37). Our previous study with an ␣ 4 mutant, Y991A, unable to bind paxillin, indicated a major defect in both cytoskeletal anchoring and ␣ 4 adhesiveness measured under shear stress conditions. Anchorage was proposed to protect adhesion receptors from membranal uprooting (38). However, we could not observe any shear force-driven pulling of this mutated integrin subunit from the membrane even when it was occupied by high affinity mAb. 3 Instead, the contribution of integrin anchorage to bond stabilization could be 2-fold: first, to allow the integrin-ligand complex to instantaneously load low forces and undergo in situ bond strengthening (4), and second, to disperse the loaded forces along elastic regions of the leukocyte via microvilli extension (39). In support of the first mechanism, a recent computer simulation of the ␣ v ␤ 3 -ligand complex suggests that the integrin headpiece can be instantaneously stabilized by force application on the ligand-integrin complex (3). Furthermore, the VLA-4-related ␣ 4 ␤ 7 integrin favors a critical shear stress to tether on ligand (40). In light of its selective contribution to VLA-4 adhesiveness under shear stress, it is possible that paxillin enables ␣ 4 integrins to properly load forces and undergo instantaneous ligand-induced stabilization. Studies from the selectin field suggest that the elasticity of leukocyte microvilli also stabilizes adhesions by dispersing forces along the microvillar axis (41). Paxillin ␣ 4 linkages may thus optimize force dispersion of both VLA-4-VCAM-1 and VLA-4-mAb interactions along the microvillar axis. Both microvilli-expressed selectins and ␣ 4 integrins must be prop-erly anchored to the cytoskeleton of these microvilli in order for their bonds to benefit from such a force dispersion mechanism (39).
Talin 1, unlike paxillin, is key for the generation of the major high affinity VLA-4 subset in Jurkat T cells. Assessing the role of this subset in adhesive VLA-4 functions, we found that high affinity VLA-4 is preferentially implicated in immediate stabilization of VLA-4-mediated cell arrests on both medium and high density VCAM-1 but is dispensable for weaker rolling and transient adhesions. Nevertheless, when VCAM-1 becomes limiting, Jurkat T cell attachments, although transient, are mediated exclusively by a talin 1-dependent high affinity VLA-4 subset. Therefore, the more abundant talin 1-independent VLA-4 subsets can operate without the high affinity VLA-4 subset to support lymphocyte attachments to medium and high density VCAM-1 but not to low density ligand. Thus, talin 1-stabilized high affinity VLA-4 subsets expressed on circulating effector T cells may have a key function in the firm adhesion of these cells to vascular beds expressing low density VCAM-1. Although we could not detect a similar high affinity VLA-4 subset on primary lymphocytes, talin 1 played a role in these cells similar to that played in Jurkat. Since our analytical system was based on single siRNA constructs and luciferase controls, we cannot formally rule out the possibility of off target effects of the silencing procedure employed. However, we carefully confirmed the specificity of both constructs by verifying that they do not affect the expression or function of all relevant cell surface proteins functionally characterized. Furthermore, siRNA to paxillin did not affect the expression of talin, and conversely siRNA to talin 1 did not affect the expression of paxillin. We have also ruled out any effect of these constructs on key cytoskeleton regulatory proteins, since neither of them affected cell chemotaxis or L-selectin adhesiveness under various shear stress conditions, a readout that requires both cytoskeletal integrity and conserved microvillar structure.
Taken together, our results suggest that ␤ 1 associations with talin 1, the exclusive talin member expressed in T cells, and of ␣ 4 with paxillin independently regulate the affinity and the mechanical stability of the VLA-4-VCAM-1 bond. These distinct roles for paxillin and talin 1 in VLA-4 adhesiveness to VCAM-1 are shared, however, by both Jurkat cells, which express VLA-4 in high affinity conformation and by resting T lymphocytes, which express VLA-4 in a much lower affinity conformation. Distinct ␣ and ␤ subunit partners cooperatively regulate the subunits to which they bind in other integrin het-3 E. Manevich and Ronen Alon, unpublished results. erodimers. For instance, replacement of the ␣ subunit tail of the aIIb ␤ 3 with that of the ␣ 5 tail, alters the activation state of the chimeric integrin while retaining the dependence of this activation on the ␤ 3 adaptor, talin (8). Furthermore, previous studies conducted on ␣ 4 ␤ 1 transfected into either myeloid lymphoblastic K562 cells or epithelial cells (Chinese hamster ovary) showed that substitutions of ␣ 4 tails with other ␣ subunit tails altered VLA-4 adhesiveness without changing integrin affinity states, which were most likely regulated by talin (19,42).
How does talin 1 up-regulate VLA-4 affinity in resting cultured Jurkat T cells? Recently, the RIAM adaptor, a profilin ligand that binds activated Rap1, was shown to activate the GPIIb␤ 3 integrin by recruiting talin to the integrin tail (10). RIAM overexpression in Jurkat cells was correlated with increased spreading and adhesiveness to fibronectin under shear free conditions (43), and RIAM suppression was shown to abrogate TCR-mediated integrin activation (44). Thus, it is possible that this adaptor may also contribute to rapid VLA-4 adhesiveness to VCAM-1 by recruiting talin to the vicinity of VLA-4. The talin head is composed mainly of a FERM (band four-point-one, ezrin, radixin, moesin) domain with binding sites for the cytoplasmic tails of integrin ␤ subunits (45), focal adhesion kinase, and actin (46). Talin can be activated by binding phosphatidylinositol 4,5-bisphosphate (PIP2) both prior to and following integrin occupancy with ligands (47). PIP2 exposes the integrin-binding site on talin and increases talin affinity to ␤ tails, including that of ␤ 1 (47). Small pools of this phosphoinositide exist on the plasma membrane and may be enriched on surface microvilli, where VLA-4 is also enriched (48). Since talin serves as a scaffold for the PIP2 generating phosphatidylinositol phosphate kinase type I␥ (49), preformed VLA-4-talin complexes may be specifically enriched with this enzyme and with its product, PIP2, which in turn, drives talin to stabilize subsets of VLA-4 at high affinity (28). Consistent with this possibility, sequestering PIP2 from the surface of Jurkat T cells with a gelsolin-derived cell-permeable peptide impaired VLA-4-mediated attachments to VCAM-1, although an exclusive role for talin 1 in this PIP2-dependent VLA-4 function has not yet been confirmed. 4 Although talin binding to ␤ 1 integrins has been shown to regulate focal adhesions of multiple ␤ 1 integrin members in fibroblasts (6,50) and was implicated in both the ligation and the tethering of ligand-occupied integrins to the cytoskeleton (33,51), our study is the first to show a role for talin 1 in VLA-4 affinity regulation in general and in hematopoietic cells in particular. Talin 1 is unlikely to be the exclusive cytoskeletal modulator of VLA-4 affinity. Threonine 788, for instance, although implicated in ␤ 1 integrin activation, is not required for talin head binding to the ␤ 1 subunit tail (52) and thus must operate through an as yet unidentified ␤ 1 adaptor. Talin binds a membrane-proximal NPXY motif on the ␤ 1 tail (53), and mutations that disrupt this motif inhibit both talin binding and integrin activation (54,55). A recent paper indicated that deleting the membrane-proximal motif, NPIY, in the ␤ 1 tail of VLA-4 expressed in Jurkat T cells results in impaired arrest on VCAM-1 under shear flow (56), but the role of talin-␤ 1 associations and of high affinity VLA-4 subsets have not been reported. Our results are consistent with a role for this NPIY site in the talin 1-generated high affinity VLA-4 subset that mediates immediate T cell arrest on VCAM-1 under shear flow. Strikingly, although chemokines stimulate inside-out conformational changes in the LFA-1 integrin prior to ICAM-1 binding as detected by multiple extension and ␤-I domain reporters (26, 57), we could not detect any analogous conformational changes in the ␤ 1 integrins of PBLs or Jurkat cells. We therefore conclude that chemokines can stimulate VLA-4 adhesiveness in T cells without switching it in situ to a high affinity conformation. Rather, we hypothesize a role for chemokine signals in stabilizing VLA-4 occupied by surface-bound VCAM-1 under shear stress conditions (i.e. an outside-in integrin activation process). These results contrast the ability of certain chemokines and chemoattractants to stimulate through inside-out signaling VLA-4 affinity to soluble VCAM-1 and conformationally switch ␤ 1 integrins into highly active states in monocyte lines and primary monocytes (58). Nevertheless, in the absence of such inside-out conformational switches of T cell VLA-4, talin 1 can still contribute to this outside-in activation of VLA-4 by chemokine, possibly by enriching a priori the number of high affinity VLA-4 available for chemokine stimulation. Recent results from our laboratory also suggest that these events are favored by external forces applied on the VLA-4-VCAM-1 bond. 5 Future studies should therefore aim at identifying pathways utilized by distinct chemokine signals to recruit talin 1 at nascent adhesive contacts generated under either shear-free or shear stress conditions (10,59). We anticipate that the adhesive potential of both resting and chemokine-4 E. Manevich and R. Alon, unpublished observations. 5 E. Woolf and R. Alon, unpublished observations.